Breast Cancer Classification Using Transfer Learning with the MobileNetV2 Architecture
Year: 2026
Type: article
Abstract: Breast cancer is still one of the most commonly diagnosed cancers in the world, so detecting it early and reliably matters a great deal for patient outcomes. This paper looks at how machine learning and deep learning methods can tell benign tumors apart from malignant ones, using structured clinical and diagnostic attributes instead of raw imaging data. The model is trained and tested on patient records that include age, tumor size, lymph-node involvement and prior diagnostic history. We describe the preprocessing pipeline step by step: how the categorical diagnosis labels are encoded, how non-informative identifier fields are dropped, and how the numerical features are standardized before training. We also run an exploratory analysis of the dataset, looking at the balance between diagnosis classes, patient age groups, tumor size and breast side, so the classification results can be read against a clear picture of the data. Traditional machine learning classifiers are then compared with deep learning approaches from recent literature, and our results are discussed alongside the limitations reported by other studies. Throughout, we present the model as a decision-support tool that complements established clinical diagnosis and does not replace it.
Author: Kotramma Mathada
Language: English
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7235209523
